Skip to content
APMIX.AI

One API key for every top AI model.

Point Claude Code, Codex CLI or your own code at api.apmix.ai. GPT, Claude, Gemini, Grok, DeepSeek and Qwen run on a single key and one monthly plan.

  • OpenAI- and Anthropic-compatible endpoints
  • Works with Claude Code, Codex CLI and Cursor
  • Cancel any time
Free3M free tokens on 1 model at sign-up
~/app — zsh
claude
ANTHROPIC_BASE_URL=https://api.apmix.ai
ANTHROPIC_AUTH_TOKEN=apx_live_••••d41e
claude
Claude Code · sonnet-5.5 · apmix
Rate-limit /login and add tests
Read src/routes/login.ts
Edit src/routes/login.ts
+38 −4
Write tests/login.test.ts
+61
Done · 2 files · 48k tokens
codex
OPENAI_BASE_URL=https://api.apmix.ai/v1
OPENAI_API_KEY=apx_live_••••d41e
codex "soft-delete invoices"
Codex · gpt-6-luna · apmix
Edit db/schema.ts
+12 −2
Write db/migrations/0042.sql
+19
Run pnpm test 42 passed
Done · 2 files · 31k tokens

50 models from 12 families, on one API key.

All models

Keep your tools. Change two settings.

Claude Code, Codex, Cursor and the official SDKs all read a base URL and an API key. Point them at apmix, and every model in your plan answers on the same key.

Step-by-step guides

Anthropic format
export ANTHROPIC_BASE_URL="https://api.apmix.ai"
export ANTHROPIC_AUTH_TOKEN="apx_live_YOUR_KEY"

For Claude Code, the Anthropic SDKs and any client of the Messages API.

OpenAI format
export OPENAI_BASE_URL="https://api.apmix.ai/v1"
export OPENAI_API_KEY="apx_live_YOUR_KEY"

For Codex, Cursor, OpenCode, Kilo Code, the OpenAI SDKs and any client of Chat Completions or Responses.

Replace apx_live_YOUR_KEY with a key from your dashboard. Each guide has the exact file and setting for its tool, and the model to start with.

One account instead of one per provider.

No separate accounts, keys and invoices for each AI lab. This is what you get instead.

One key

The same key calls every model in your plan. To switch models, change the model name in the request.

deepseek-v4.1-flash · glm-5.3 · gemini-4-argon

Both API formats

OpenAI's Chat Completions and Responses, and Anthropic's Messages, with streaming and tool calls. Clients built for either one work unchanged.

/v1/chat/completions · /v1/responses · /v1/messages

One allowance

Each plan has a monthly allowance of weighted tokens, shared by all your models and keys. There are no daily or weekly caps unless you set them.

Starter 60M · Pro 200M · Max 500M

One bill

A fixed monthly price, paid by card, Apple Pay or Google Pay. If a month runs short, a one-time top-up covers it and never expires.

Starter $4.99/month · Pro $14.99/month · Max $29.99/month

What a month of each plan buys.

Plans come with a monthly allowance of weighted tokens. Each model has a rate, set by what it costs to run, and every token you send or receive counts at that rate.

At 0.5×, 1,000 tokens use 500 weighted tokens. At 4×, the same 1,000 use 4,000.

Rates for every model
The models apmix serves most, and how many tokens a month of each plan buys on them.
ModelRateStarter60M a monthPro200M a monthMax500M a month
DeepSeek V4.1 Flashdeepseek-v4.1-flash0.5×120M400M1B
GLM-5.3glm-5.31.5×Not included in Starter133.3M333.3M
Gemini 4 Argongemini-4-argon3.5×Not included in Starter57.1M142.9M
GPT-6 Astragpt-6-astra10×Not included in StarterNot included in Pro50M
Claude Opus 5.5claude-opus-5-54×Not included in StarterNot included in Pro125M
GLM-5.3 Flashglm-5.3-flash0.3×200M666.7M1.7B

Input, output and reasoning tokens all count. Each plan includes the models of its own tier and of every plan below it.

Questions

Something else? Write to support@apmix.ai.

Does it work with Claude Code and Codex CLI?

Yes. Set the Anthropic or OpenAI base URL to api.apmix.ai and use your apmix key. Cursor, official SDKs and curl work the same way.

Do I need separate accounts with each provider?

No. One apmix.ai key covers every model in your plan.

Can I use the same key for every model?

Yes. Change the model name in the request; the key and the endpoint stay the same.

What are weighted tokens?

One unit across every model. Each model has its own multiplier, and every token you send or receive costs that many weighted tokens: 1,000 tokens are 500 weighted tokens on a 0.5× model and 4,000 on a 4× model. One allowance covers all of them; the pricing page lists every model's rate.

What happens when I run out of tokens?

Requests return a clear error until the monthly reset, or you move to a bigger plan or buy a one-time usage top-up. Both apply right away.

How much does it cost, and is it cheaper than paying each provider?

Plans start at $4.99 a month for 60M weighted tokens; bigger plans add more models, up to the whole catalog — GPT, Claude, Gemini, Grok, DeepSeek and Qwen — on one key. That is one cheap subscription instead of six pay-as-you-go bills.

How do I cancel?

From your account page, in one click. Your keys keep working until the end of the paid month.

Start with any plan. Change it whenever you like.

Choose your allowance, create a key, and point your tools at apmix.